My app is having tearing problems when I scroll fast horizontally. Normally I'd just enable vsync, but I don't know how to do that in Qt. The tearing is occurring on my dev PC which runs Linux with a X-server, but it will also need to work well on the target system which is either Android or Linux+EGLFS.
One google result was from the Qt4 era and said that the tearing will be gone in Qt 5. But apparently it isn't.
Many google results are about env vars for embedded linux systems, such as QT_QPA_EGLFS_FORCEVSYNC, but I'd prefer the solution to work for my dev PC as well.
One google result said that vsync being enabled/disabled depends on the renderloop in use. I don't override the renderloop, so, since I'm on linux, I think I'm using the threaded renderloop.
Here's a bugreport about nvidia binary drivers causing this.
https://bugreports.qt.io/browse/QTBUG-45480
I too am using those drivers, so this may be it - just a driver bug. If so, at least it's unlikely to happen on my target hardware (I still haven't tested there), so that's good.
Related
As an Android developer I've been moving away from Eclipse to Intellij IDEA for production code in anticipation of Google's Android studios which shares a code base with IDEA.
My experience has been a good one up to this point. I've only been using IDEA at the office, where I have a 4x core Intel i7 machine running Ubuntu 12.04 LTS (Sun JDK/JRE), up to this point and I've never noticed what the performance of IDEA really is.
Now however after setting IDEA up on my personal computer at home the performance is abysmal. Memory usage is normal, but the constant CPU usage bounces between 80%-100% (over the whole application lifecycle). And that is when nothing else is running on the machine and no work is being done, by me or visually by the IDE.
This makes IDEA unusable when working on it, and I can forget about having anything else running along side it.
My home specs and software are:
Intel Core 2 duo 3GHz
8 GB RAM
Ubuntu 12.04 x64 LTS (3.8.0-35-generic) running of SSD SATA
Intellij IDEA 13.0-0ubuntu1 build: IC-133.193
Tried both OpenJDK and Sun
And the strange thing is that this happens as well with Android Studios.
All help in trying to debug this behaviour would be appreciated.
#Edit 1:
Noticed that the CPU load falls down to 20% when bringing up dialogs (Project structure, Settings, etc) and then goes right back up when dismissing them.
#Edit 2:
I tested simply getting the tarball straight from JetBrains, instead of using the one in Canonical's ppa. The performance was significantly better for at least an hour (20-30% CPU usage while idle). Seems that the native file watcher in C-PPA wasn't working properly and was indexing the whole filesystem.
However the performance became worse after the first hour or so, going back to 90-100% CPU.
The issue turned out to be the native-file watcher being out-of-date. IntelliJ was re indexing my whole drive it seems. Was fixed by uninstalling the version gotten from Canonical's ppa and installing directly from JetBrain's own webpage.
Are you using any plugins outside of the included ones which might cause issues.
I don't run Ubuntu anymore but can't recall any issues with high CPU-usage when i did. (I use Fedora with KDE a colleague uses Fedora with GNOME though. )
Does this always happen or only when you have a project open?
I'm thinking if this might have something to do with the background-compile that IDEA does.
Might be worth trying to turn this off.
Found under Project Settings -> Compiler -> Make Project Automatically
worst case it is a Unity-integration issue or something. Haven't used unity so can't say.
Usualy I manage to fix it by deleting IDE's index files rm -rf ~/.RubyMine60/system,
don't forget to change .RubyMine60 to IDEA's config folder
If you're willing to do some sleuthing, you could run the Oracle JVM and use the VisualVM profiler to see where the IDE is spending all its time, presuming it's a Java-based process that's actually eating your CPU cycles.
I have an application using ThreadX 5.1 as the kernel.
The Image is flashed on to a hardware running an ARM 9 processor.
I'm trying to build a Simulator for the application that can be run on Windows (say XP, 32-bit).
Is there any way I can make it run on Windows, without modifying the entire source code to start calling win32 system calls?
You can build a Simulator for the application that can be run on Windows with "ThreadX for Win32".
"ThreadX for Win32"'s specification is hear.
http://rtos.com/products/threadx/Win32
Yes you can if you are willing to put in the work.
First observe that each threadx system call has an equivalent posix call except for events.
So your threadx program can run as a single process using posix threads, mutexes, etc.
Events can be handled by an external library (there are a few out there).
If you find this difficult in windows then the simplest thing to do is set up a linux vm. I use an ubuntu vm running on Virtual Box. It is very easy to set up. All you will need is the cdt version of eclipse.
Next you need to stub out all of your low level system calls.
This is also easier than you might think. For example, if you have a SPI driver to read and write to flash, you can replace your flash with a big array which is quite easy to work with at this level.
Having said all this, you may get more mileage if your threadx application is modular. Then you can test each module on it's own and you don't need to mess with threads, etc.
As a first approximation this may give you what you need without going the distance to port the whole thing to run under posix.
I have done this successfully in the past and developed a full set of unit tests for a module that allowed me to develop and test it (on my mac) before going to the target. Development is much faster and reliable this way.
Another option you may want to consider is to find a qemu project that supports your microprocessor. With some work you can develop a complete simulator for your platform and then run the real firmware under the emulator.
Good luck.
Using NDK r8c, Eclipse 4.2, Windows 7 64
I've used remote debuggers before (on other platforms, via gigabit ethernet) for large C++ codebases that felt no different than local debugging. The Java debugger that comes with the SDK runs fast too. Therefore I'm quite baffled why gdb is so slow to connect and step over lines of code.
In my current application, which is around 20 static libraries and 1500 source files, it takes about 15 seconds to connect, and about 2 seconds to step. I'm more concerned about stepping.
Has anyone ever profiled gdb to see what the problem is? If so, any suggestions?
I have. My cohorts and I at NVIDIA have contributed several commits to AOSP to address this problem, although our focus has been on shared libraries (symbol load performance, and pending symbol resolution.) We have sped up solib load processing by a factor of 6x. (Although, after doing our own work we discovered that 3x of that 6x had already been solved upstream by GNU, in 7.5... so we abandoned our reinvention, and submitted the relevant 7.5 patches up to Google's NDK repository, which was based on the older 7.3 GDB.) I believe all of our speedups are present in r8d... but I haven't checked.
I cannot think of any reason why static libraries would slow things down, but I must admit I haven't given any thought to them. Do you have a specific reason for believing so, or was that just comment to give perspective about the size and scope of your debugging needs?
We have begun to work on the stepping problem, but don't have anything to share yet. Basically, the bottleneck is ADB (especially on Windows.) Additionally, there is a lot of chatty communication between GDB and gdbserver, when stepping, especially if you are using an IDE with local window, register window, expression window, stack window, etc., all updating with each step. That's a lot of chatter that could likely be optimized for the IDE use-case.
Just some of the fixes that we are considering for speeding up stepping will be IDE-specific:
Using python scripting to pre-process watch expressions in GDB, rather than in the IDE.
Implementing "super-packets" communicating between GDB and gdbserver... packets that encapsulate IDE-specific communications in a way that minimizes chatter between GDB and gdbserver.
We intend to share all of this with the Android community.
I'm looking for a really really small linux distribution or process of making my own that's sole purpose is to get an air application to launch full screen and stay there; Essentially I'm building a home kitchen computer that runs entirely as an AIR app.
I have looked into using windows xp; and windows xp embedded but they pose so many issues I figured I'd try modern linux.
I have also seen TinyCore Linux which looks interestingly small but not sure what issues that poses in regards to running AIR and "hardware" accelerated display. I've also thought about stripping down an Ubuntu installation but I'm sure somebody must have done this already; google is just failing me right now...
I'm also interested in running an "embedded" version of say android and running the air app on some arm-based hardware again; with just the AIR runtimes only - although this is less preferred as it's more complex.
I'm also hooking this up to a touch screen monitor (not yet arrived) so I'll need to hunt down or write some drivers for translating the touch events into something AIR can understand... (this was my main intention for using windows in that all the drivers will just work).
What I'm after
Minified Linux kernel with JUST the drivers for the box I need
X Display with accelerated graphics support (Doesn't have to be X if AIR can run on a frame buffer?)
Running a Full screen AIR application (simple enough)
Ability to write back to the filesystem (enough support for AIR)
SSH Access for remote control
Samba for updating the filesystem (easier to maintain the system)
Touch screen support (3M Ex III I think...)
Audio support
Don't need
Don't need any window manager or any other GUI tools unless required by AIR
Don't need any toolbars or file managers or anything; The AIR app is the "OS"
Don't need any package managers or repos
Don't need multi user or logging in; everything can just run as an unprivileged account
Don't need to
I don't mind hand crafting the filesystem and configs if that makes it easier; I'm mainly looking for a "filesystem" that is as tiny as possible that I can just plop my AIR app into and write some scripts to get it to start when the X server starts
Thanks,
Chris
Try an embedded Linux build system such as Buildroot. It can build an entire system from source, and be very lightweight. The basic system is less than 1 MB in size.
Ended up going with Tiny Core. Very tiny and quick to boot up. You can also write extensions for it and you don't have a persistent drive which allows you to just switch the thing off without worry that it's going to break something -- exactly what you need in a kitchen :-D.
My current plan is to:
Just set up a working version using Ubuntu as this is mostly supported by Adobe
Slowly strip it back and try and get as little things to start as possible on boot
Try building my own distro/package from source and selecting only the packages I need
Compile my own kernel with nearly everything turned off and just leave on the things I need
Hopefully this still falls within StackOverflow's umbrella!
I'm looking to create a quick boot linux laptop for my wife. All it really needs is to be able to do is browse the internet (with flash and video etc.).
Are there any distros that are made for this, or any guides out there that show good ways to speed stuff up? I've read that I should "remove stuff from the kernel that I don't use" but that's a little out of my skill set.
Thanks!
If you're using Ubuntu (or a variant, like xubuntu or kubuntu), there is a package called BootUp-Manager. There's an article about it over at Lifehacker. It lets you check and uncheck things in the startup and shutdown scripts to optimize things (such as turning off checking for new hardware, or whatever)
You may also be able to gain a simple speed-up by going into System->Administration->Services and disabling any services you don't need.
If you'd like to see how much time is being spent on each part, install the package Bootchart, and that should give you a detailed profile of everything that goes on during startup, and let you focus on the most time-consuming parts, and measure your progress as you tune the system.
I believe Xubuntu is designed for low memory footprint/fast booting and whatnot while still having a decent amount of features. Not a Linux user but it just seems to stick out in my head.
Some guys got an EEE PC netbook booting in 5 seconds running a modified version of Fedora. Might be a good starting point: http://lwn.net/Articles/299483/
Try: Damn Small Linux is a very versatile 50MB mini desktop oriented Linux distribution.
Alternatively, get an Asus motherboard with expressgate - it has an onboard Linux (spashtop) that boots in 3 seconds. Its designed for quick web surfing, IM, music etc whilst still letting you boot into your main OS.
If you really want it to boot fast, I would suggest creating an initrd containing exactly the software you need to do what you want it to do. The initrd will get read from the disk once as one large file, and then everything will run out of ram.
This is not an easy solution, the easiest solution will be jishi's solution of using a Live CD, but, that won't be the fastest solution.
I have been using Kubuntu 14.04.4/5 on a Dell laptop with dual boot. Am not real happy at the moment with it. You are all correct about the slowness of a liveCD.
There are LiveCD-versions of working linux-distros with browser and installed flash, java.
Check out LiveCD
http://en.wikipedia.org/wiki/Live_CD
you will find links to different flavours with download.
There are also USB-drive-versions.